Gibson: Flora and vegetation of Middle and South Ironcap, Digger Rock and Hatter Hill 
flats =5, small rise in valley =6). Slope was scored on a 
one to three scale from flat to medium, to steep. Aspect 
was recorded as one of 16 cardinal directions. Altitude 
was taken from 1:100000 series topographical map to 
nearest 10 m. Vegetation structure was recorded using 
Muir's (1977) classification. All quadrats were 
permanently marked with four steel fence droppers and 
their positions fixed using a GPS unit. 
Quadrats were classified according to similarities in 
species composition. In these analyses only perennial 
species were used to facilitate comparisons with 
classifications from other ranges (Gibson el al. 1997; 
Gibson & Lyons 1998a,b; Gibson & Lyons 2001a,b). The 
quadrat and species classifications undertaken used the 
Czekanowski similarity coefficient and "unweighted 
pair-group mean average" fusion method (UPGMA 
module in PATN, Belbin 1995, beta value -0.1, Sneath & 
Sokal, 1973). Semi-strong hybrid (SSH in PATN) 
ordination of the quadrat data was undertaken to show 
spatial relationships between quadrat groups (here 
referred to as community types) and to elucidate possible 
environmental correlates with the classification (Belbin 
1991). Methods of Dufrene & Legendre (1997) were used 
to determine best indicator taxa for each group (from PC- 
ORD v 4.24, McCune & Mefford 1999). 
Climate estimates (mean annual temperature, annual 
temperature range, mean annual rainfall, rainfall 
coefficient of variation) were obtained from BIOCLIM 
(Busby 1986), a prediction system that uses 
mathematical surfaces fitted to long term climate data. 
Relationships among and between physical site 
parameters and climate estimates was examined using 
Spearman rank correlation coefficient. To reduce the 
probability of type I errors given the number of 
intercorrelations, significance differences were reported 
at a level of P<0.01. Vectors for the physical site 
parameters, latitude, altitude and climatic estimates 
were fitted to the ordination along axes of highest 
correlation using the principal axis correlation routine 
in the PATN package (Belbin 1995) (also known as 
rotational correlation analysis). Statistical significance of 
these vectors was determined using random 
permutations of the values of the variable among sites 
(Faith & Norris 1989). Statistical relationships between 
quadrat groups for physical site parameters and climate 
estimates were tested using Kruskal-Wallis non- 
parametric analysis of variance (Siegel 1956). 
Nomenclature generally follows Paczkowska and 
Chapman (2000). Voucher specimens have been be 
lodged in the Western Australian Herbarium. Introduced 
taxa are indicated by a 
Results 
Flora 
A total of 343 taxa (species, subspecies, varieties) and 
two hybrids were recorded from the Forrestania 
greenstone belt. The flora list was compiled from taxa 
found in the 38 quadrats or the adjacent area and from 
collections of the Western Australian Herbarium 
(Appendix 1). Of these 345 taxa, 342 are native and 3 are 
weeds. The best represented families were the Myrtaceae 
(77 taxa), Proteaceae (38 taxa), Mimosaceae (25 taxa), 
Papilionaceae and Orchidaceae (20 taxa), Asteraceae (17 
native taxa and 1 introduced taxon) and Epacridaceae (14 
taxa) (Appendix 1). This pattern is typical of the flora of 
the South Western Botanical Province (Beard 1990). The 
most common genera were Eucalyptus (38 taxa), Acacia 
(25 taxa) and Melaleuca (22 taxa). 
Thirty-five taxa of conservation significance were 
recorded from the range. This included; 
• three taxa listed as threatened; 
• a further 29 that are being considered for listing as 
threatened flora (Atkins 2001); 
• 10 taxa considered endemic to the range; and 
• a further eight that are regional endemics (found 
within 100 km) (Table 1). 
One taxon ( Slenanthemum liberum) was collected for 
the first time and is only known from two populations 
(Rye 2001). 
During the current survey, new populations of Boronia 
revoluta and Banksia sphaerocarpa var. dolichostyla (both 
listed as threatened) were located. This was somewhat 
surprising given the proximity to active mines and the 
botanical survey work undertaken associated with their 
commissioning. A significant range extensions was 
recorded for Bentleya diminuta. This is a very unusual 
tufted perennial herb (Pittosporaceae) which has small 
clusters of prostrate leaves connected by thick 
underground rhizomes. It has previously been recorded 
from the Cape Arid-Ravensthorpe area. 
Vegetation 
In the 38 quadrats established south of Middle 
Ironcap, 229 taxa were recorded of which 202 were 
perennial. Fifty-seven perennials occurred at only one 
quadrat. Preliminary analyses showed these singletons 
had no effect on the community classification and 
therefore are not discussed further. As a result the final 
data set consisted of 145 perennial taxa in 38 quadrats. 
Species richness ranged from seven to 36 taxa per 
quadrat, with individual taxa occurring in between two 
and 28 of the 38 quadrats. Only material that could be 
identified down to species or subspecies level was 
included in the analysis (ca 95% of records). 
The first major division in the dendrogram separates 
the quadrats on skeletal soils derived from banded 
ironstone and the massive laterites (community types 1 
& 2) from quadrats on deeper soils derived from 
greenstone or decomposing laterites (community types 3 
& 4; Fig 2, Appendix 2). 
• Community type 1 is comprised of the species-rich 
shrublands or mallee shrublands. Species in 
species groups I and K characterize community 
type 1 and contain most indicator species 
(Appendix 2). Average species richness was 27.2 
taxa plot 1 . This community type was restricted to 
the massive outcrops along the range (Middle 
Ironcap, South Ironcap, Digger Rock and Hatter 
Hill, Fig 1). Three subtypes can be recognized 
(Appendix 2). Type la occurred on all outcrops 
and was strongly represented by species group I 
and a lack of species group J which characterized 
51 
